Pancreas list - dcp to tier 1

[arschat]

Lucia reached out to us to help with tier 1 of pancreas.

The pancreas team has had some troubles progressing their work in the last few months for a whole bunch of reasons. To move forward with the blessing of the bionetwork, they are manually collecting their tier 1 metadata (as much as possible) for publications.

Is there any way they could easily leverage the wrangling you have already done to get the partially filled metadata sheets to data contributors to fill in the gaps?

Using the DCP to Tier 1 notebook, we can generate tier 1 spreadsheets.

[arschat]

We have successfully generated spreadsheets for 17 out of 20 projects in the pancreas list here using the notebooks mentioned.
However, after a check there were some inconsistencies on the donor number (2 projects), and most of the projects had accessions as IDs. This would most definately not match with contributor's tier 1, so we requested from integration leads to handle any potential Tier 1 IDs existing in the matrices, and then we could do the mapping of IDs using the names.

list of projects & inconsistencies

project name	problem	percent of required Tier 1
CPmicroEnvironment	inconsistent donor list (12 in publ, 34 in dcp)	76
dcp5924379466506983	inconsistent donor list (25 in dbgap, 39 in dcp)	62
EastAsianPancreaticIslets	accession for donor_id, sample_id, cs_id (GSE97655)	69
Faryabi-Human-10x3pv2	accession for cs_id, ID in name	72
GarciaOcana-Human-10x3pv3	accession for sample_id, cs_id	69
Healthy_and_type_2_diabetes_pancreas		#N/A
HealthyAndDiabeticPancreas	accession for sample_id, cs_id (GSE101207), different donor_id than figure 2A	66
Herrera-Human-10x3pv3	accession for donor_id, sample_id, cs_id (GSE150724)	69
Herrera-Human-10x3pv3	accession for cs_id (GSM2194190)	69
HumanEndocrinePancreas	generic cs_id (GSE81608)	72
HumanIsletType2Diabetes		72
HumanMousePancreas	prefix in donor_id, s. accession in cs_id, id in name	62
HumanT2DPancreas	accession for sample_id, cs_id (GSE198623)	72
Lickert-Human-10x3pv2	accession for sample_id, cs_id (GSE85241), id in name	69
pancreasCelSeq2	accession for sample_id, cs_id (GSE114297)	72
pancreasNormalIslets	accession for sample_id, cs_id (GSE183568), id in GEO name & CS name	76
scHumanPancreaticIslets	accession for cs_id (GSE81547), id in GEO name	69

[arschat]

Waiting from Sara Jimènez and Shrey Parikh to reply with concatenated datasets and the appropriate IDs.

[arschat]

Sara and Shrey provided us the contacting sheet they use.

Project Name	Project UUID	Study	DOI	DCP	Ingest
CPmicroEnvironment	c5ca43aa-3b2b-4216-8eb3-f57adcbc99a1	GSE165045	10.1136/gutjnl-2021-324546	DCP	Ingest
dcp5924379466506983	bcdf233f-9246-4c0c-9843-0514120b7e3a	GSE155698	10.1038/s43018-020-00121-4	DCP	Ingest
EastAsianPancreaticIslets	e77fed30-959d-4fad-bc15-a0a5a85c21d2	GSE97655	10.1038/s41598-017-05266-4	DCP	Ingest
Faryabi-Human-10x3pv2	daef3fda-2620-45ae-a3f7-1613814a35bf	GSE148073	10.1038/s42255-022-00531-x	DCP	Ingest
GarciaOcana-Human-10x3pv3	17cf943b-e247-454f-908b-da58665fcc56	GSE217837	10.1186/s13073-023-01179-2	DCP	Ingest
Healthy_and_type_2_diabetes_pancreas	ae71be1d-ddd8-4feb-9bed-24c3ddb6e1ad	E-MTAB-5061	10.1016/j.cmet.2016.08.020	DCP	Ingest
HealthyAndDiabeticPancreas	1c6a960d-52ac-44ea-b728-a59c7ab9dc8e	GSE101207	10.1016/j.celrep.2019.02.043	DCP	Ingest
Herrera-Human-10x3pv3	7a8d45f1-353b-4508-8e89-65a96785b167	GSE150724	10.1038/s41467-022-29588-8	DCP	Ingest
HumanEndocrinePancreas	99101928-d9b1-4aaf-b759-e97958ac7403	GSE83139	10.2337/db16-0405	DCP	Ingest
HumanIsletType2Diabetes	7adede6a-0ab7-45e6-9b67-ffe7466bec1f	GSE81608	10.1016/j.cmet.2016.08.018	DCP	Ingest
HumanMousePancreas	f86f1ab4-1fbb-4510-ae35-3ffd752d4dfc	GSE84133	10.1016/j.cels.2016.08.011	DCP	Ingest
HumanT2DPancreas	c6ad8f9b-d26a-4811-b2ba-93d487978446	GSE86473	10.1101/gr.212720.116	DCP	Ingest
Lickert-Human-10x3pv2	28dd1438-8f40-40d0-8e53-ee3301b66218	GSE198623	10.1016/j.molmet.2022.101595	DCP	Ingest
pancreasCelSeq2	894ae6ac-5b48-41a8-a72f-315a9b60a62e	GSE85241	10.1016/j.cels.2016.09.002	DCP	Ingest
pancreasNormalIslets	27e2e0ae-5971-4927-aac1-19e81804097b	GSE114297	10.2337/db18-0365	DCP	Ingest
scHumanPancreaticIslets	daa371e8-1ec3-43ef-924f-896d901eab6f	GSE183568	10.1172/jci.insight.151621	DCP	Ingest
SPAN	1e618693-aa16-4414-bb11-5c8baf9a5c4d	PANC-DB	10.1038/s42255-023-00806-x	DCP	Ingest
HPAPCorrPatil10x	b3938158-4e8d-4fdb-9e13-9e94270dde16	Charite snRNA	10.1053/j.gastro.2020.11.010	DCP	Ingest
Chen-Human-ATACseq	de0bb4b0-b691-46d7-83db-55670f0afc3a	GSE233476	10.21203/rs.3.rs-3343318/v1	DCP	Ingest

[arschat]

• sample_source since all datasets did not have the newly implemented transplant_organ field filled, I investigated separately the source of datasets. All projects mention or imply that pancreas donor was originated for transplantation, or similar procedure was executed.

  specifically

  • The Prodo laboratories, specify that they...

  deliver human research pancreases from United Network for Organ Sharing (UNOS) from listed cadaver organ donors that are refused for primary human pancreas transplantation or isolated islet transplanted into listed diabetic recipients.

  • Integrated Islet Distribution Program (IIDP) specify ...

  The IIDP depends on the subcontracted human islet isolation centers to provide research investigators with human islets. It is the responsiblity of the human islet centers to obtain research quality pancreata from the Organ Procurement Organizations (OPO). Some criteria are stricter than those used by transplant centers for organ transplant donors.

  • Alberta Diabetes Institute IsletCore specify in their record sheet that they're following a procedure that includes cold perfusion
  • For project CPmicroEnvironment

  Pancreatic tissue was only used if the pancreas could not be used for clinical pancreas or islet transplantation

[arschat]

Draft email.

Hi everyone,

After our discussion last week, I validated and mapped ~75% of required tier 1 fields and some available optional. The folder with the files is here https://drive.google.com/drive/folders/1iTioXQelz3_GEyZ4UlcOkUYyKmHk7eq1 .

Filename pancreas_dcp_tier1.xlsx has all the Tier 1 metadata that we currently hold in DCP as tier 1 metadata, mapped to the library level. The most useful information is in tab dcp_tier1_metadata and other tabs are there to validate between what you've already wrangled in "contacting_sheet", and our metadata.

There are a few comments I would like to make:

1. I included some extra fields that are not part of Tier 1 in order to help validation and mapping of IDs. These fields have a ~ as prefix.
2. For manner_of_death we did not have any information on hardy scale, however, we usually have some info about the cause of death which might be feasible to translate to hardy scale (field ~manner_of_death_string).
3. For sample_source since in most cases the organism was refered as organ donor/ donor organism or any kind of cold perfusion is referenced. Therefore, I assumed that in all cases the donor was not postmortem donor but organ donor or surgical donor (biopsy or blood draw).
4. sample_collection_method for organ donation I added surgical resection, however I am not sure if that would be consistent across bionetworks.
5. tissue_type is tissue, however it is standard procedure for almost all projects to isolate pancreatic islets and culture for a few hours/days. Existing CxG datasets that do follow this procedure have tissue in the tissue_type and not cell culture or organoid therefore I kept the same for all other datasets.
6. sequencing_platform there is one dataset (PANC-DB) that two sequencers are mentioned but it is not clear which library was sequenced with which sequencer.
7. library_preparation_batch and library_sequencing_run was not recorded at the time of dataset wrangling however accessions or any other information recorded is provided in case it might be useful.
8. is_primary_data is not information that we can record in our schema.
9. reference_genome for dataset GSE198623, in your contacting_sheet is listed as GRCh38 however in GEO it is mentioned that it's hg19/ GRCh37
10. disease_ontology_term_id there is one dataset (PANC-DB) where the presence of Auto-Antibody (AAB) is mentioned in the disease field, however, as far as I understand, that presence does not constitute disease or a MONDO ontology term. Therefore, we have listed these donors as normal/ non-diseased.
11. self_reported_ethnicity_ontology_term_id for Tier 1 as far as I am aware this is going to be filled with "unknown" due to privacy reasons. However, I've added the ethnicities that we currrently hold in DCP here (recorded from public archives).
12. There are 2 published datasets that are yet to be wrangled.

I also added mappings for the IDs you provided in the all_ids_pancreas_wo_index.csv file. Here are some comments about those:

1. There are 5 datasets that have one singular donor_id, however we've wrangled multiple donors (GSE217837, GSE81608, GSE83139, GSE165045, GSE183568).
2. There are 79 sample GSM* ids that match to a mouse sample. Would we like us to include mouse data as well? There are some projects from the list that have both human and mouse data.
3. There are IDs from Escape which is not yet published. We are not able to wrangle and provide any metadata for projects that are not published yet.

I hope this will help you with your integration steps. Let us know if you have more questions.

Best regards,
Arsenios

[arschat]

Added also the charite dataset, and asked question about further wrangling:

Hi Sara, Daniel and Shrey,
I wanted to give you an small update about this. We added the charite snRNA dataset in the pancreas_dcp_tier1.xlsx file.
There is another dataset GSE233476 colored "peach" in the contacting_sheet - contacting, that we can't find single cell data to download, only bulk RNA & ATAC seq. We were wondering if you had the chance to get the single cell data from contributors, or if you are still insterested in the bulk part of dataset as well.
In the all_ids_pancreas_wo_index file, we found some IDs from the nicheformer dataset. However, neither nicheformer, nor the original publication of those samples ( GSE156728) were in the list of the contacting_sheet. Would you like to include any of these in the publication list as well?
All pancreas datasets that are marked as green or peach in contacting_sheet have been wrangled, converted to Tier 1 format and appended in the pancreas_dcp_tier1.xlsx file, except from the GSE233476 mentioned above and the unpublished onces which we don't have access to.
Let us know if that's enough or if we can help with something else.
Best,
Arsenios